CSS for book descriptions?

[mirage]

Does one of the CSS files in /data or elsewhere govern the rendering of text on the Description screen (that contains the descriptive metadata visible on the metadata screen in Calibre and in the epub's opf file) that's accessible from, among other places, the Book Information screen for each book? Mainly, I want to create space between paragraphs so that multi-paragraph descriptions don't all look like one paragraph. I can do this by manually adding one space paragraphs in Calibre or in the opf directly, but it would be nice not to have to bother with that.

[Galunid]

https://github.com/koreader/koreader....lua#L194-L205
https://github.com/koreader/koreader...ua#L1002-L1054

So basically it replaces paragraphs/br tags with new line, and strips the rest. No CSS involved at all. You can probably tinker with those two functions in util.lua, to get what you want, but that'd require you to build your own .apk.

[mirage]

Thanks for pointing the way. I'm using KOReader on a Kobo device, not an Android one, FWIW. But that probably doesn't matter if I'm changing the code and compiling as long as I do it for Kobo. But I don't expect to do that. I was hoping for a CSS solution to this.

[Frenzie]

That particular bit of code is ~4 years old, so you could also try to rejig it with the newer HTML widget. Could you upload a scrambled sample document that demonstrates the problem?

On Kobo you don't need to build anything to play around with the code btw. For most things like this you can just edit the Lua files directly on the device itself. But be careful, they will be overwritten if you update so you'd better make sure to keep a copy if you do.

[mirage]

Originally Posted by Frenzie

That particular bit of code is ~4 years old, so you could also try to rejig it with the newer HTML widget. Could you upload a scrambled sample document that demonstrates the problem?

On Kobo you don't need to build anything to play around with the code btw. For most things like this you can just edit the Lua files directly on the device itself. But be careful, they will be overwritten if you update so you'd better make sure to keep a copy if you do.

OK. Thanks. I should have realized that there's no need to compile since I realized that the lua files are present on the device in that form. I can play with that code.

There's probably no need for the scrambled epub. Any epub with an opf file with multiple paragraphs of description in the <dc:description> tag will do. This is not a bug or something specific to a particular book. It's universal.

Look at what KOReader displays when you bring up that description text the way I mentioned above. There is no spacing between paragraphs or indentation of the first line. In other words, all the paragraphs vertically render so that it looks like there is only one long paragraph. That's why I was hoping to use css properties like margin to create some distance between paragraphs, at least as long as the styles didn't affect anything in the text of the books themselves. The Nickel reader seems to do this by whatever means it uses. I can get around this in KOReader by putting a paragraph containing only a space

<p> </p>

between all paragraphs with text in the opf for the aforementioned tag. I do this in Calibre, but it probably doesn't matter if I do it there or in Sigil or whatever text editor. That renders vertical space between the paragraphs with content, but I have to go to the trouble of doing it and then the Nickel reader renders the same thing with far too much space because it already handles that rendering with plenty of space to begin with. That's not exactly a catastrophic problem, particularly since I mostly read in KOReader. But why not obsess over it if I can?

[Frenzie]

Originally Posted by mirage

There's probably no need for the scrambled epub. Any epub with an opf file with multiple paragraphs of description in the <dc:description> tag will do. This is not a bug or something specific to a particular book. It's universal.

Whether or not there's a need for a scrambled EPUB, there's a need to not make us waste time trying to find or create an EPUB that displays the relevant properties.

[mirage]

I've attached a scrambled epub.

Alas, ScrambleEbook removed the description text I'm writing about from the opf. So I replaced it with some Lorem Ipsum stuff. The effect is the same as what I'd mentioned earlier. You'll see that when displaying Book Information > Description that there's no vertical spacing between paragraphs of the text. My hope was to be able to use CSS to create that space.

[mirage]

Perhaps making it easier for everyone, I took pictures of the screen I'm referring to.

In picture 1, you can see that there's space between the paragraphs. I put that into the metadata in the Calibre "Edit metadata" window by executing a CR in the description text field between each paragraph which causes Calibre to insert a <p> tag with no content other than a space in each such case.

Picture 2 shows the way it renders when each paragraph (defined by what's between <p> tags) is not separated by that extra CR/empty <p>.

I want it to look like picture 1 without having to edit the html of the metadata to achieve that. I don't mind doing the edits, but that causes excess vertical spacing in the Nickel reader because the latter provides spacing between paragraphs to begin with.

I can provide samples of anything else that you might like to see with regard to this, too.

[Frenzie]

I only took a quick peek, but as @Galunid proposed perhaps the to plain text algorithm would behave more as expected like this (certainly a lot quicker than putting in an HTML widget):

diff --git a/frontend/util.lua b/frontend/util.lua
index 863ce849..6293e5c6 100644
--- a/frontend/util.lua
+++ b/frontend/util.lua
@@ -1002,8 +1002,8 @@ This may fail on complex HTML (with styles, scripts, comments), but should be fi
 function util.htmlToPlainText(text)
     -- Replace <br> and <p> with \n
     text = text:gsub("%s*<%s*br%s*/?>%s*", "\n") -- <br> and <br/>
-    text = text:gsub("%s*<%s*p%s*>%s*", "\n") -- <p>
-    text = text:gsub("%s*</%s*p%s*>%s*", "\n") -- </p>
+    text = text:gsub("%s*<%s*p%s*>%s*", "\n\n") -- <p>
+    text = text:gsub("%s*</%s*p%s*>%s*", "\n\n") -- </p>
     text = text:gsub("%s*<%s*p%s*/>%s*", "\n") -- standalone <p/>
     -- Remove all HTML tags
     text = text:gsub("<[^>]*>", "")

[mirage]

[Frenzie]

Discussion in https://github.com/koreader/koreader/pull/7027